Luxeon power supply

[star882]

http://www.e-insite.net/ednmag/contents/images/217668f1.pdf
Has anyone tried this circuit to power a luxeon?

 

[MrAl]

Hello,

I, for one, wouldnt use two op amps just
to sense current to the LED. Apparently
the designer didnt know how to achieve both
the required gain and a differential input
with only one op amp section and associated
resistors.
Also, sensing near ground is sometimes better
as the high end common mode input range usually
isnt there so you end up needing a much higher
battery voltage.
Also,
most LED users want to be able to power
their light off of batteries, so the
15v supply voltage wouldnt be available,
nor would the input requirement of 7v.
National makes op amps that work down to
1.8 volts or something, so that might be
a better idea, or else use one of the
Zetex current sense chips. These are both
nice small packages too.

Good luck with your LED circuits,
Al

 

[georges80]

I figured some of you might be interested in the following boards I made. Hopefully this isn't some breach of protocol on this list - if so, please ignore this posting.

I have a bunch of PCBs that I've had made that can either drive the Luxeon 1W unit or with some component changes (pin compatible) the 5W unit.

For a 1W white Luxeon with a Vf of 3.1V the driver board operated down to 3.8V before the Luxeon drive current went out of regulation (less than 350mA). The board is 1" square - bolts to the back of a Luxeon Star or Star/O. It can operate up to 30V, the efficiecy is about 72% at 12.8V and 86% at 4.8V (4 AA NiMH). The design is constant current - using a low value sense resistor - schematics are on my web site. It is a step down design - since it's primary function is to drive a luxeon for a dome light replacement for a vehicle (12V).

You can check out my web site for more details:
http://www.geocities.com/george_tlc/led.html

The PCBs have been made for some fellow 4wd owners to retrofit their dome lights. I did a run of 200 PCBs to get the price down and I have about 100 boards beyond my needs. I call sell (within the US) for $3 each + postage. If anyone is interested check out my website & you can find my email address there. They are blank boards, you can get some of the components (IC's, diodes & inductor) as samples and the rest are mostly jelly bean R's & C's - all surface mount.

If there's enough interest, I could 'bulk' buy the R's & C's and provide those for about $3.

cheers,
george.

 

[georges80]

Jon, the advantage of the LM7301 op amp is you can get free samples - it also has a very wide supply range - I have a 24V diesel 4wd that this would also work in.

Yes, for the 5W Luxeon you use the pin compatible LM2675 (1A). You also need to change the inductor (slightly taller version) for higher current. A shorter version is for the 350mA output (same footprint) - inductors are also free samples from coilcraft.

The two schottky diodes are free samples from On Semiconductor.

You also have to change a feedback resistor to change the current limit to 700mA instead of 350mA.

I optimized for 12V and higher, but it will work fine at lower current - just overkill in that situation. You can see from the efficiency curves that I measured, it works quite well.

There are more efficient switchers out there (high freq 1Mhz etc), but they don't work much above 6V input voltage - important for my application.

NS has nomographs to show how to choose inductor & cap size for various input/output voltage configurations.

My overall goal was to build a rugged, efficient, but Cheap (samples where possible) design. I also wanted true constant current regulation.

cheers,
george.

 

[Lucien]

*cough* This might not be the best time or place to ask this, but has anyone taken a look at the MAX1627?

It claims high efficiency and low drop out voltage, maybe not so important in automotive use, but useful perhaps for a hand-held.

Max supply voltage is 16.5V, another automotive limitation, and it looks like it won't power the 5W.

George,
Very nice work

Are all your designs step down?

 

[Jonathan]

George,

I took a look at the PCBs that you had made, and they are _quite_ interesting. I've been looking at using the 'simple switcher' in a current regulating design myself.

There is a 'dealers corner' on CPF, and raw commercial announcements should go there. On the other hand, answering a question with a 'I designed and am selling a board that does just that!' strikes me as entirely appropriate, and I've done it myself with other circuits. But since this is a discussion area, I feel that it is fair game to ask about the circuit being suggested.

I was wondering if you'd considered using a current sense feedback chip such as the Zetex ZXCT1010? These chips are specifically designed to amplify the voltage across a low value sense resistor and provide a higher voltage to a normally voltage regulating device. They essentially fill the role that you are using the op-amp for, but they are tweaked for the particular application.

Additionally, the LM2674 is only rated to 500mA...do you use a different device for the 700mA output needed for the 5W devices?

Finally, I'd appreciate some advice on how you selected your inductor value. Did you optimize for 12V operation or for lower or higher voltage?

Thanks
Jon

 

[Jonathan]

The MAX1627 _should_ be able to power the 5W with an appropriate external transistor. I'd guess the big issue with the 1627 is that it does require the external pass transistor. The 2674 and 2675 have the pass transistor built in, which makes for a simpler circuit but generally means greater losses in the transistor.

Have you considered the MAX1744, which operates at higher input voltage, and includes a current sense circuit? It appears that the current sense is intended to limit the maximum switch current, but I think that it could be used to regulate the output current to the load.

-Jon

 

[Lucien]

Originally posted by Jonathan:

Have you considered the MAX1744, which operates at higher input voltage, and includes a current sense circuit?

<font size="2" face="Verdana, Arial">Not untill just now

. The 1744 does seem better, but I don't actually understand about the current sense circuit.

Actually, there's a lot about these circuits I don't understand. I'm still starting out, so I hope you could expain some of this stuff to me.

I never thought of being able to use the MAX1627 (and the 1744 now) for 5W units as I always thought the upper limit was 5V. Reading through the datasheet closer, I realise that that was for the 1626... doh!

As for regulating the output current to the load, I think using an op-amp based circuit with a current sense resistor would be best.Using the current sense pin would probably be far more tricky and provide poorer regulation. And at this point, with the external transistor, the LM based circuits would be easier, so all this is kinda reinventing the wheel. Doh again!

But if we're talking battery powered, I think the 1627 has the advantage in terms of low dropout voltage.

 

[georges80]

Seems like most of you are somewhat stuck on Maxim parts - I don't find their selection that great.

Linear Technology has a very broad selection of switchers - especially high frequency units over 1MHz. Their advantage is very small capacitors and inductors (can you say TINY!) and highly efficient. For example the LT3401 runs over 1Mhz, is capable of providing 500mA at 3.3V from 1.8V input. It'll run a 1W LS, but not a 5W. Check out their range of stuff. http://www.linear.com

My switchers for LEDS have generally been down converters, mainly because I run my Luxeons either from 12V or 4 AA NiMH and not particularly 'tiny' since I have plenty of room in my applications.

Regarding current regulation, setting the switching current is not the same as setting a true current limit. You would have to tweak the switch current to get close to the required value on a per unit basis, especially given the variability in Vf of some Luxeons.

What we really need is to have Maxim or LTC come out with a switcher that uses a lot less that 1.2V for the FB pin. i.e. they need to put the op-amp 'inside'. Though the Zetex 1010 is pretty cheap - just takes more room...

It's hard to design a single circuit that will step up or down or both with very wide input voltage range and efficient at the extremes and tiny and....

Guess what I'm trying to say (I design lot's of switching supplies where I work - for digital cameras running off batteries) that you really need to come up with the intended use and optimize for that, i.e. how many batteries, what type of battery, what output voltage/current and required battery life and amount of real estate available and cost target. Then you have a hope of narrowing down an optimized solution.

In the case of my NS based down switcher, the goals in order of importance: automotive 12V input, lowest possible cost (IC's, diodes & inductor as samples), bolt on to back of Luxeon Star or Star/O, capable of 4 AA NiMH operation and finally efficiency.

I think I achieved most of that, cost of parts $3 or less (getting samples as above) + board + luxeon. The bonus is that it can drive the 5W Luxeon with some component changes (pin compatible).

But, there's more than one way to skin a cat, and someone can most likely come up with a different design & beat me out on some of the goals...

george.

 

[Jonathan]

George,

I personally feel that you did a rather nice job meeting your design goals.

My goals are slightly different, in that I don't want to be limited by sample quantities, and want to select parts that are least expensive when ordered through common sources, eg. Digikey. But I've been looking for a good design to run the 5W LS devices from a 12V supply, and I have to say that your's is quite nice.

I am actually most familiar with Linear Tech, as I used the LT1618 to build a boost converter for running LEDs. The 1618 has two feedback inputs: a 1.25V referenced input for voltage regulation, and a 50mV differential referenced input for current regulation. In my application, the voltage feedback is set to limit the output voltage in case the output goes open circuit, and the primary regulation is based on the current through a small value sense resistor. This chip is perfect for running strings of smaller LEDs, but is too small for running the fat LS LEDs.

WRT the MAX1744, the current sense looks like it is intended to be used to limit switch current, which as you said is very different from regulating the output current. However I think that if you were to connect the current sense resistor and the voltage feedback _after_ the output capacitor, then the current sense would act to regulate the output current, essentially disabling the switch if the current sense voltage were to exceed 100mV. Alas it would appear that Digikey doesn't have stock and the maxim direct sales page is down....

 

[georges80]

Originally posted by Jonathan:
[WRT the MAX1744, the current sense looks like it is intended to be used to limit switch current, which as you said is very different from regulating the output current. However I think that if you were to connect the current sense resistor and the voltage feedback _after_ the output capacitor, then the current sense would act to regulate the output current, essentially disabling the switch if the current sense voltage were to exceed 100mV. Alas it would appear that Digikey doesn't have stock and the maxim direct sales page is down....[/QB]

<font size="2" face="Verdana, Arial">Jon, I agree, if you wire the max1744 up the way you describe it appears that it should current limit based on current draw by the load. The only downside of the 100mV fixed sense voltage is that for such small resistors (under 1 ohms) you can't easily get them in fine resistance increments. 0.27 ohms would give 370mA drive and 0.15 ohms 667mA. Good enough if you aren't striving to exactness of a particular overdrive level.

cheers, off camping this w/end, with LEDs lighting the way

george.